Parallel algorithms for variational inequalities over the Cartesian product of the intersections of the fixed point sets of nonexpansive mappings (Q953825): Difference between revisions

Let \(H_1,H_2,\dots,H_n\) be Hilbert spaces and \(H=H_1\times H_2\times\dots \times H_n\) their Cartesian product space equipped with the inner product \[ \langle \langle \mathbf{x},\mathbf{y}\rangle\rangle=\sum_{i=1}^{m}\langle x_i,y_i\rangle,\quad\forall\;\mathbf{x}=(x_1,\dots,x_m),\;\mathbf{y}=(y_1,\dots,y_m)\in H. \] Denote by \(\| \cdot\| _i\) the induced norm on \(H_i\), \(i=1,\dots,m\) and by \(\| | \cdot| \| \) the induced norm on the Hilbert space \((H,\langle \langle \cdot \rangle\rangle)\). On \(H\) we consider a second norm, defined by \[ \| \mathbf{x}\| _{\infty}:=\max_{1\leq i \leq m} \frac{\| x_i\| _i}{w_i},\quad\forall\;\mathbf{x}=(x_1,\dots,x_m)\in H, \] called weighted maximum norm, with the positive weights \(w_i,\;i=1,\dots,m\). A mapping \(T: H\to H\) is said to be block-nonexpansive if \[ \| T\mathbf{x}-T\mathbf{y}\| _{\infty}\leq \| \mathbf{x}-\mathbf{y}\| _{\infty},\quad\forall\;\mathbf{x},\mathbf{y} \in H. \] The authors consider parallel iterative algorithms for a family of block-nonexpansive mappings \(T_i:H\to H\) and prove that the sequence generated by this algorithm converges strongly to the unique solution of a certain variational inequality.